I will discuss first the case of a force exerted between two objects by a cable being reeled in.
From there I will move on the the case of gravity.
Let it be granted that $F=ma$ holds good.
And I need to make this explicit: let it be granted that $F=ma$ will obtain the same at every location of space, and independent of orientation in space.
Let there be two objects, in space, small enough that any gravitational acceleration is negligable. We name these object 1 and object 2.
(Other than that a large population of objects is present, with gravitational acceleration between these objects negligable, thus the population as a whole provides reference of acceleration for the motions of object 1 and object 2.)
Let a cable be stretched between object 1 and object 2. For the benefit of the thought demonstration we treat that cable as massless. One of the objects starts reeling in that cable.
The following sequence is observed:
State A: first the two are both stationary
They start moving towards each other
When the two come in contact the two stick together and in that final state (state B) they don't have a velocity with respect to State A.
The following is of central importance: the resulting motion does not tell us which object was reeling in the cable.
This is because in space the only leverage you have is your own inertial mass. By contrast: on Earth you can attach yourself to the Earth surface, thus securing yourself to something much more massive than yourself; in space the only leverage that you have is your own inertial mass.
In space the amount of leverage that any object has (for the purpose of reeling in another object) is given by F=ma. The amount of opposition to being reeled in is also given by F=ma
Is the process of reeling in mutual? Or is it one-sided? The point is: that is indistinguishable; on the basis of the resulting motion we cannot eliminate either possibility.
Since we cannot eliminate either possibility we turn that around and make it an element of our theory of motion that no such distinction can be made.
So we have the following:
If it is granted that F=ma will obtain in the same way at every location in space:
For motion with respect to the Common Center Of Mass (CCOM) of object 1 and object 2:
$$ m_1 * a_1 = m_2 * -a_2$$
It follows that during the entire time:
$$ m_1 * v_1 = m_2 * -v_2 $$
It follows that with respect to the CCOM the total linear momentum remains zero.
And of course Newton's third law follows logically:
$$ F_{reaction} = -F_{action} $$
Gravity
We infer the existence of gravity from how gravity affects motion. It is the resulting motion that is observable, not gravity itself.
So: what happens if the following question is raised: 'is gravity one-sided, or is it mutual?'. The point is: there is no way to addres that question. Either way the resulting motion is the same.
Discussion:
Philosophy of physics
Quote from physics.stackexchange contributor knzhou
[...] in physics, you can often run derivations in both directions: you can use X to derive Y, and also Y to derive X. That isn't circular reasoning, because the real support for X (or Y) isn't that it can be derived from Y (or X), but that it is supported by some experimental data D. This two-way derivation then tells you that if you have data D supporting X (or Y), then it also supports Y (or X).
So: we need to keep in mind that derivation does not have an intrinsic direction.
Another way of saying this:
In any logical system there is great freedom to exchange axiom and theorem without changing the content of the system.
That said: I have a preference for the sequence as presented above:
The summary:
Take as starting point a combination of $F=ma$, and assertion of uniformity. The assertion of uniformity: $F=ma$ will obtain the same at every location in space.
Those two assertions are sufficient to imply conservation of momentum and Newton's third law.